Phylogenetic linear modeling results

Settings

This report was generated automatically by phylogenize (Bradley and Pollard, submitted: https://phylogenize.org), with the following parameters:

• Type of input data: “16S”
• Phenotype calculated: “specificity”
• Environment of interest: “Plant rhizosphere”
• Prior probabilities for environments: “uninformative”

Phenotype calculation

Below, trees for all phyla with at least 5 representatives in the input data are presented. Trees are colored according to the phenotype chosen (here, specificity in the environment Plant rhizosphere).

For prevalence, the tree is colored black (low prevalence) to orange (high prevalence). For specificity, the tree is colored red for taxa that are specific for an environment, blue for taxa that are specific for other environments (i.e., found less often in the environment chosen), and gray for taxa that are neither.

Phenotype mapped across trees

All taxa

Only mapped/observed taxa

Displays phenotype on trees, with only the mapped/observed taxa plotted.

Associations between genes and taxa

This table gives an overview of how many genes were significantly associated at an FDR of 5%.

	significant genes
cyanobacteria	0
firmicutes	2029
actinobacteria	1715
proteobacteria	2880
acidobacteria	0
bacteroidetes	472
verrucomicrobia	4
delta.epsilon.subdivisions	393
planctomycetes	0
spirochaetes	0
fusobacteria	6

Gene association heat maps

The left shows the phenotype painted on tips of the tree as above. The right adds a heatmap where blue points represent presence of a gene (aligned with the tips of the tree) and white points represent absence. Only significantly-associated genes are displayed. Genes are clustered from left to right by Jaccard dissimilarity, so genes with similar patterns of inheritance will form vertical stripes. Horizontal stripes may be visible for taxa that have particularly extreme phenotypes.

Enrichment analysis

This process takes the set of genes that are significant and positively associated with the phenotype, then asks whether there are more than expected by chance in a given SEED subsystem. Subsystems come in three levels of granularity (referred to here as l1, l2, and l3); “l1” subsystems are the most general and “l3” are the most specific.

All SEED categories returned are significant at an FDR of 25%. “Strong”, “Med”, and “Weak” refer to per-gene significance cutoffs of 5%, 10%, and 25%, respectively.

##             used  (Mb) gc trigger   (Mb)  max used   (Mb)
## Ncells   5074385 271.1   15233987  813.6  15233987  813.6
## Vcells 121451501 926.7  327878196 2501.6 327878196 2501.6

##             used  (Mb) gc trigger   (Mb)  max used   (Mb)
## Ncells   5059895 270.3   15233987  813.6  15233987  813.6
## Vcells 120091059 916.3  327878196 2501.6 327878196 2501.6

Full output files

Links to other files generated by the report:

• phenotype.tab
• all-results.csv
• pos-sig-thresholded.csv
• enr-table.csv
• enr-overlaps.csv
• enr-overlaps-sorted.csv